For effective validation of computational fluid dynamics (CFD) codes for design, numerical simulations must be compared and contrasted with experimental data sets which are well posed and measured. This paper presents, with the emphasis placed primarily upon the description and explanation of the experimental results, a test case designed for CFD validation concerning the flow field generated by the presence of tip clearance in an annular cascade turbine blade row.The test case is based upon a moderately loaded prismatic blade profile, operating with axial inflow and a nominal mid-span exit Mach number of 0.5. In addition to the case with no clearance, three gap heights, representing 1, 3 and 5 per cent of chord length have been selected. These tip clearances were chosen because they represent three conditions where the tip clearance is below, approximately equal to and substantially larger than the inlet boundary layer displacement thickness. At the leading edge this produces different behaviours, with the smallest gap producing a stagnation point, equivalent to the case without tip clearance, while the largest clearance allows throughflow over the leading edge into the tip gap.In the presence of larger tip clearances, the tip leakage flow induces a stagnating flow in the trailing-edge region, which results in a large area of blocked flow in the blade passage downstream of the trailing edge.

Recent discussions in the industrial CFD community have identified a need for guidelines covering the accurate and efficient computation of a range of flow field classes. This paper addresses some of these issues for a standard turbomachinery test case, by investigating the flow through on annular blade row of a generic turbine profile, operating at an exit Mach number of 0.5. The joint experimental and CFD works have focused upon identifying and quantifying the loss sources and loss development. This has been achieved by the acquisition of dense data sets of a known, high and repeatable experimental accuracy, where, concentrating primarily upon the investigation of the secondary flow phenomena, optimised experimental methods have been employed to measure the pressure distributions in the annuls and the development of the flow field, particularly the loss structures, downstream of the trailing edge. On the CFD side, the flow field has been computed using commercial codes. Adopting the loss distribution as a primary marker for the quality of the CFD results, the performance and efficacy of the codes and the implemented viscous models can be assessed. The flow has been computed both 2D and 3D, from inviscid to laminar to turbulent with different turbulence models, with and without transition. According to the model, the flow has been investigated considering a wide range of parameters influencing its turbulent state. Through this study, guidelines concerning numerical smoothing and free-stream turbulence parameters are proposed for the computation of such flows. The need of a transition model within 3D schemes, rather than an improvement of the turbulence model, to predict accurate loss levels has been recognized. However, through the cross analysis of the different computational results, a good estimate of the loss magnitude and distribution is feasible with the currently used models.

A numerical analysis using a commercial unsteady Navier-Stokes solver has been performed on a pin/blade configuration, in order to assess the efficacy of a commercial code in calculating time-periodic interactions and to gain a better understanding of the unsteady flow physics in axial turbines. Two cases have been investigated, with the pin positioned at 25% and 50% of true chord ahead of the leading edge. Both configurations have been computed both 2D and 3D. The 2D case was used to examine the influence of numerical parameters, such as mesh, time and space discretisation. The 3D case allowed insight into the complete flow field including the wake influence on the secondary flow and mixing processes of the blade row. The basic mechanisms of the wake-blade interaction proved, as expected, to be the same for both pin positions. Yet, as the closest pin wake interaction with the blade field was much stronger, its features have helped to identify the respective roles of wake fluid transport and blade potential field for both cases. The latter effect, noticeably strong with the thick leading edge blade form presented in this study, has been often neglected, and this study helps shed new light on this phenomenon. The code used had been validated in previous work for pin-free steady flow within the same blade row and the new time dependent case has served to confirm the code range and limitations.

A numerical analysis using a commercial unsteady Navier–Stokes solver has been performed on a pin/blade configuration, in order to assess the efficacy of a commercial code in calculating time-periodic interactions and to gain a better understanding of the unsteady flow physics in axial turbines. Two cases have been investigated, with the pin positioned at 25 and 50 percent of true chord ahead of the leading edge. Both configurations have been computed both two and three dimensionally. The two-dimensional case was used to examine the influence of numerical parameters, such as mesh, time, and space discretization. The three-dimensional case allowed insight into the complete flow field including the wake influence on the secondary flow and mixing processes of the blade row. The basic mechanisms of the wake–blade interaction proved, as expected, to be the same for both pin positions. Yet, as the closest pin wake interaction with the blade field was much stronger, its features have helped to identify the respective roles of wake fluid transport and blade potential field for both cases. The latter effect, noticeably strong with the thick leading edge blade form presented in this study, has often been neglected, and this study helps shed new light on this phenomenon. The code used had been validated in previous work for pin-free steady flow within the same blade row and the new time-dependent case has served to confirm the code range and limitations. [S0889-504X(00)02104-8]

No abstract